Accounting for model parameter uncertainty provides more robust projections of dissolved organic carbon dynamics to aid drinking water management

Changes in climate and human behaviour impact catchment hydrology and the export of nutrients including dissolved organic carbon (DOC), with consequences for drinking water supply. In this study, we projected future river discharge and DOC dynamics under three Shared Socioeconomic Pathways (i.e., different futures of climatic conditions, socio-economic development and adaptation to climate change) and quantified change relative to a baseline for two contrasting catchments: one in Sweden and one in Ireland. For this, we used the Generalised Watershed Loading Functions Model (GWLF) with an integrated DOC module (GWLF-DOC) and drove it with data from an ensemble of global climate models, taking into account variability derived from multiple model parameter sets. We assessed the relative contribution of each of these two factors (climate input data and model parameterisation) to the total uncertainty in predictions. Projections for river discharge differed between the two sites in magnitude, variability and direction of change depending on the future scenario and time period. In contrast, DOC was always projected to show increases in concentration throughout the annual cycle and over time, with the highest levels by the end of the century, for scenarios with greater warming and low mitigation efforts. Future climate data provided the dominant source of uncertainty in all of our projections. However, the DOC model parameters, which respond to temperature and soil moisture conditions, became more influential in scenarios of higher climatic variability. Our approach highlights the benefits of incorporating often ignored parameter uncertainty in climate change impact assessments for both interpreting outputs and communicating results to water managers.